1408 CHAPTER 42. MAXIMAL MONOTONE OPERATORS, HILBERT SPACE

Now the first term in 42.9.65 is bounded so consider the second. The integral in this termis of the form∫ t

0e

sλ φ (s,z(s))ds+

∫ t

0e

sλ (φ (t,z(s))−φ (s,z(s)))ds. (42.9.66)

Since [z] ∈ D(Φ) , φ (s,z(s))< ∞ for all s and also the first integral in 42.9.66 is finite. By42.9.59, the second term in 42.9.66 is dominated by

∫ t

0K(

1+φ (s,z(s))+ |z(s)|2)|t− s|ds < ∞.

This shows φ (t,Jλ [z] (t))< ∞ for all t and so Φ([Jλ [z]]) is given by the top line of 42.9.61.Therefore, by convexity of φ (t, ·) and Jensen’s inequality,

Φ([Jλ [z]]) =∫ T

(t,e

−tλ x0 +

e−tλ

∫ t

0e

sλ z(s)ds

)dt

≤∫ T

0

(e−tλ φ (t,x0)+

e−tλ

∫ t

0e

sλ φ (t,z(s))ds

)dt

=∫ T

0e−tλ φ (t,x0)dt +

∫ T

0

e−tλ

∫ t

0e

sλ φ (s,z(s))dsdt

+∫ T

0

e−tλ

∫ t

0e

sλ (φ (t,z(s))−φ (s,z(s)))dsdt. (42.9.67)

By 42.9.59, the last term is dominated by∫ T

0

∫ t

0

e−(t−s)

λ K(

1+φ (s,z(s))+ |z(s)|2)|t− s|dsdt =

∫ T

0

∫ T

se−(t−s)

λ

t− sλ

dtK(

1+φ (s,z(s))+ |z(s)|2)

ds

≤Cλ +Cλ

(Φ([z])+ |[z]|2

). (42.9.68)

for some constant, C. From 42.9.59, φ (t,x0) is bounded and so the first term in 42.9.67is dominated by an expression of the form Cλ . Now consider the middle term of 42.9.67.Since φ is nonnegative,∫ T

0

e−tλ

∫ t

0e

sλ φ (s,z(s))dsdt =

∫ T

0

∫ T

s

e−(t−s)

λ dtφ (s,z(s))ds

≤∫ T

0

∫∞

0e−uduφ (s,z(s))ds = Φ([z]) . (42.9.69)

It followsΦ([Jλ [z]])≤Φ([z])+Cλ +Cλ

(Φ([z])+ |[z]|2

).

1408 CHAPTER 42. MAXIMAL MONOTONE OPERATORS, HILBERT SPACENow the first term in 42.9.65 is bounded so consider the second. The integral in this termis of the form[ eto s,2(s))ao+ [oe (6 (t,2(s)) — 0 (s,2(s))) ds. (42.9.6)JO JOSince [z] € D(®), @ (s,z(s)) < © for all s and also the first integral in 42.9.66 is finite. By42.9.59, the second term in 42.9.66 is dominated bycx [ K(1+0(s.2(9)) +9?) It —s|ds <0,This shows @ (¢, J, [z] (t)) < c¢ for all t and so ® (J, [z]]) is given by the top line of 42.9.61.Therefore, by convexity of @ (t,-) and Jensen’s inequality,® ((Jy [z]]) =['o (netaa+ 67 [ <\as) dt<[" (<4 6 (0.10) + 56% [ eo (e(9)4s) dt-[ aa) (tx0)dt+ f "Ley [eto z(s)) dsdtXr 0 ,+f zx [ ex (@ (t,z(s)) — 0 (s,z(s))) dsdt. (42.9.67)By 42.9.59, the last term is dominated byLhefle(1 +06, z(s )) + l2(s)?) |t —s|dsdt =* Sark (1 + (s,z(s))+ iz(s)/") ds<CA+CA (® ([z)) + Il ) . (42.9.68)for some constant, C. From 42.9.59, @ (t,x9) is bounded and so the first term in 42.9.67is dominated by an expression of the form CA. Now consider the middle term of 42.9.67.Since @ is nonnegative,[ zet [ bo tscls)) )) dsdt = [ [ge = ato ( (s,z(s)) ds< [ [ e “dud (s,z(s))ds = ®([z]). (42.9.69)It follows® (Ja [e]) < P(e) +CA +2 (() +I).